Project-MONAI / Project-MONAI/tutorials

Too large `step_size` in deepedit

Open
#1,430 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Jupyter Notebook
Stars
2.5k
Forks
803
Avg merge
6d 22h
Merged PRs (30d)
3

Description

Describe the bug
Too large step_size which may cause the learning rate never decreases unless the model is trained for 5000 epochs.
https://github.com/Project-MONAI/tutorials/blob/bbc4e180f5130859396a98e35523bc73fa694595/deepedit/ignite/train.py#L268

Expected behavior
Set smaller step_size or set epoch_level=False in LrScheduleHandler

Originally posted in https://github.com/Project-MONAI/MONAI/issues/6625

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Open deepedit/ignite/train.py around line 268 and inspect the LrScheduleHandler configuration, especially step_size and epoch_level. Confirm how the current setting affects learning-rate decay, then adjust the scheduler configuration so decay occurs during the intended training range and verify the tutorial behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
jupyter-notebook, python, pytorch
Domain
machine-learning
Issue type
Bug
Difficulty
2/5
Estimated time
1-3 hours
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
55/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.